On disproportionate impact

There are enormous equity issues involved in COVID-19 – in exposures, outcomes (including non-COVID health and mental health), social and economic implications, and resource allocation. It matters how we construct the indicators we use to monitor performance, if we want to be able to proactively address emerging inequities. One resource I have appreciated is the Kaiser Family Foundation’s analysis of racial and ethnic disparities – it visualizes the percentages of vaccinations, cases, deaths, and total population which a given racial or ethnic group accounts for in each state. Providing these four measures together offers insight into which groups may be over- or under-represented among vaccinated individuals, considering that group’s share of the population or the burden of COVID that community has borne. Here in Massachusetts, the disproportionately low rate of vaccination among people of Hispanic ethnicities is striking – which multiple surveys suggest cannot be explained fully by individual hesitancy, pointing to issues of access and structural barriers at play.

One other question this visualization raises is why Hispanic individuals are over-represented among COVID cases to a greater degree than among COVID deaths. One explanation is that the age structure of Hispanic populations in Massachusetts tends to skew younger than White populations, so deaths due to COVID among Hispanic individuals are diminished somewhat simply because most mortality is occurring among older people – and there are more older people who are White. However, the story is more complicated.

It is already established that COVID mortality rates are significantly higher among Black individuals in the U.S. compared to White individuals – and, if we disaggregate with more detail, we also see higher mortality among certain racial and ethnic groups, such as those that fall under the over-broad categories of Asian and Hispanic. But, if we look age-adjusted rates, or drill down to age-specific rates, it becomes clear that the disproportionate mortality burden is even worse – in short, younger people of color are dying of COVID at rates far higher than would be expected. 

The COVID Community Data Lab published a research brief in December that visualizes this phenomenon in Massachusetts. The interactive charts toggle between crude COVID mortality rates among Asian, Black, Hispanic, and White residents, and age-adjusted rates. It shows that the age-adjusted mortality rate for Hispanic residents is almost 3.5 times higher than the crude rate, and around 3 times higher than the age-adjusted mortality rate for White residents. Mortality rates among Asian and Black individuals increase sharply after adjusting for age as well – while the rate actually decreases for White individuals. The National Center for Health Statistics has a tool for exploring similar disparities across states.

An October article by Harvard public health researchers looked at the national picture, and their findings suggest that age-specific mortality rates could show even starker ratios. For example, the death rate (nationally) among Hispanic individuals ages 35-44 years was 8.8 times higher than the death rate among White individuals in that same age bracket. The death rate among Black individuals ages 35-44 years was 9 times higher.

How do we interpret these findings? We know that structural racism and economic inequity shape the reality that Black and Hispanic people are more likely to hold jobs in which they are at increased risk of exposure to COVID, more likely to live in crowded households, and less likely to have access to paid sick leave – all of which, among other factors, lead to the disproportionate impact of COVID infection. We also know that the median age of the Hispanic population in Massachusetts is even younger than the Black population. This likely at least partly explains why we see the relationship we do between cases and deaths among Hispanic individuals – and why the true mortality burden is only illuminated if we examine death rates within age groups and/or remove the impact of different age structures.

Even still, there is such diversity within racial and ethnic groups, and within-group differences are obscured by these broad categories. We are probably still missing important nuances, which is why more comprehensive data collection is so important – and why qualitative assessment and sense-making, together with members of communities most deeply affected, are essential.

In my own work as a community & population health epidemiologist, there are a few things I would like to see happen. First, we need to strengthen our state’s capacity for collecting and reporting on the full set of variables required by Chapter 93, An Act addressing COVID-19 data collection and disparities in treatment. The state currently reports on race and ethnicity, gender, and age, but not on language, occupation, or disability. Second, we need better visibility into temporal and community-level trends in access, exposures, and outcomes to detect inequities and monitor how strategies are working – and we need all of this for vaccination data as well. Third, we need to address root causes: by promoting workplace protections, living wages and paid leave, safe and affordable housing, and access to health care, to start – without regard to immigration status and in ways that are linguistically and culturally accessible. Improving these socioeconomic conditions would have an outsize impact on reducing COVID cases and deaths overall – and on reducing inequities along lines of race and ethnicity.